Twitter  Is the rise of reader revenue stopping not long after it started — at the New York Times and elsewhere? nie.mn/1mZvm1C  
Nieman Journalism Lab
Pushing to the future of journalism — A project of the Nieman Foundation at Harvard

Tweet Buttons are less of a big deal than they used to be for your Twitter strategy

Whether it’s the growth of mobile, the shifting user base of Twitter, or something else, those sharing buttons appear to be generating a shrinking share of tweets linking to news stories.

tweet-buttonA year and a half ago, I wrote an article trying to gather some data on a simple question: Are all those Tweet Buttons you see pockmarking the web really useful? Are they unimportant clutter slowing page loads and uglying up your design — or are they an important tool for getting your content to spread?

A piece by the designer Oliver Reichenstein had made the useless-clutter argument; Upworthy’s Luigi Montanez argued they were more effective than some thought and wrote a script that let you check how many of recent tweets to a given site were generated by a Tweet Button. I used that script to check on a few dozen news organizations of various kinds. The result was that, for most news sites, somewhere around 1 in 5 tweets linking to their stories came from one of those buttons. Not a majority, but also not zero. Here’s the chart, from May 2012:

tweets-coming-from-tweet-button-news-orgs

And then…I forgot about it! Until Monday, when The Boston Globe’s Joel Abrams tweeted for an update:

Never let it be said I don’t take requests, because I went back to the same set of sites Tuesday and ran the numbers again. And there’s a very clear trend: As a driver of tweets, Tweet Buttons are less important than they used to be.

That doesn’t mean that they’re unimportant, or even that their use is declining. (Overall use of Twitter is up substantially over the past 18 months, after all.) But it does mean that a smaller share of tweets to news stories come from clicking that button than used to.

In the sample I looked at in May 2012, an average of 20.02 percent of tweets came from those buttons. For the new sample I took Tuesday, that number had dropped to 12.61 percent.

Here’s that same chart, but with last year’s data in red and this year’s in gold (click to expand):

tweet-button-2012-2013-grouped

Of the 37 sites surveyed, all but two saw a declining share of Tweet Button-generated tweets. (The exceptions are The Globe and Mail, which had freakishly low numbers last year for some reason, and CNN.) Here are all the news sites, ranked by share for both samplings; you can see how the decline quite clearly in aggregate:

tweet-button-2013-ordered-chart

(One technical note: Two of the sites I checked last year, Bay Citizen and California Watch, have since been subsumed into the Center for Investigative Reporting, so I used data for all of cironline.org for both in the new sample.)

All of the caveats I mentioned last year are still true; this is a sample of the last 1,000 tweets to each news site, and no doubt the numbers would be at least a little different if you ran this test at a different time of the day or a different day of the week. (Both last year’s and this year’s samples were gathered in late morning Eastern time on a weekday.) You’d want to take many more samples (and look at the mix of what else is producing tweets) to feel completely confident about your findings. (I took another sample today of the same sites, just to get another data point, and the results were virtually identical to yesterday’s: 12.89 percent vs. 12.61 percent.)

But the trend is consistent enough across all the sites to make me feel pretty good there’s something real going on here. What could that be? Let’s run through some possibilities.

People are more comfortable manually copying and pasting a link into a tweet than they used to be. Possible! But if anything, I’d wager that people who’ve come to Twitter in the past year or so are likely to be less savvy about that stuff than those who’ve been using Twitter for years.

People are consuming more of their media on smartphones, where mobile layouts often omit sharing buttons. This, I think, is a very real cause. With mobile use growing, and Twitter use being disproportionately heavily on phones, I’d wager that’s worth at least a few of the percentage points in that drop. (Remember, smartphones generally have share-on-Twitter functionality built into their operating systems these days.)

People are both discovering and sharing links more on Twitter than they used to. In other words, you used to find stories in a web browser and share them on Twitter; now you find them on Twitter in the first place. That would mean direct retweeting and “Tweet this URL” functions in Twitter apps are privileged.

News sites are deemphasizing Tweet Buttons in their layouts. That may be true in an isolated case or two, but I don’t think it’s true as a trend. If anything, Tweet Buttons feel more prominent these days.

News sites are building their own custom share-on-Twitter buttons that don’t show up as the official Twitter-supplied Tweet Button. Theoretically possible, but I spot-checked a number of sites and didn’t see any evidence of it.

I lean most toward the mobile explanation, but it’s probably a mix of factors.

So what does this all mean if you run a news site? If you really hate Tweet Buttons, it’s getting easier to justify getting rid of them.

If you’re a designer, you might have an aesthetic complaint that those knobby little roundrects ruin your clean design or distract the eye from your content. If you’re a web developer, you probably don’t like the fact that Tweet Buttons (and Facebook Like buttons, and basically all sharing buttons) slow your page load and leave you reliant on another third-party service. And if you’re a privacy advocate, you probably don’t like that those buttons let Twitter learn about what sites you’re visiting.

Those are all legitimate concerns. And knowing that button-generated tweets are shrinking as a share of overall tweets means that the social damage done by removing them is shrinking over time. You will lose social shares if you do — but the trade-off is becoming more manageable over time. People are increasingly using other methods to share your stories on Twitter.

Am I going to remove the sharing buttons from Nieman Lab? No — not now, at least. Even if only about 10 percent of tweets to our stories are coming from those buttons, I value that 10 percent! For some percentage of our audience, that button still makes it easier to them to spread our stories to friends and followers. But the data’s pushing that argument in the other direction.

                                   
What to read next
npr-one
Ken Doctor    
The new app from NPR promises to learn from what you like and serve up a never-ending buffet of news and content. What’ll be the impact on local stations?
  • https://profiles.google.com/john.g.howard/about jghoward

    My practice (and that of several colleagues) is to use bit.ly, even if there is a Tweet This button – mostly for the analytics and custom urls.

  • Joel Abrams

    Thanks Josh! I would also argue that the buttons reduce the friction of sharing. If they’re there, and they use the familiar social logos, a certain percentage of users will use them who otherwise won’t share. The friction will cause some sharing to be lost.

    I ran a version of this script on Upworthy, and it was way off the charts for sharing; another prominent news site which has a nicely-designed share menu with the share tools, and it was at 1% usage. So design does matter.

  • vint

    Another possible reasons: people are reading news on smartphones more, and use the ‘share intent’ more often? (on Android, you first choose ‘share’, and then the medium such as Twitter or G+. an uptake in ‘mobile reading’ would explain why share buttons are fewer used)

  • Dan Kaplan

    Is it possible you’re missing something? Is it possible that 2012 being an election year and 2013 not being could be responsible for some of the change? The numbers show that political sites and political news sites do better generally – and did particularly well last year. That’s especially true where the tweets come from a source (Red State) which has a strong viewpoint. And, as to the Globe and Mail, well Canada didn’t have an election last year of course. I think you are on to a trend, but it’s like skewed by this – and probably not as pronounced as you show.

  • http://www.niemanlab.org/ Joshua Benton

    I doubt it. Romney had already clinched the nomination and the fall campaign was a long ways off. The same phenomenon happened to sites that have nothing to do with politics or aren’t in this country. And aside from that, while I could understand an argument that an election season might generate more tweets overall, it doesn’t make sense that it would lead to a greater share of those tweets being generated by Tweet Buttons.

  • http://www.niemanlab.org/ Joshua Benton

    Very true — that’s what I was trying to get at in my second reason.

  • http://repost.us/ John Pettitt

    The question to ask is when those tweets happened in the life of the story. I suspect what you’ll find is the tweet button has a “patient zero” quality in that lot of share chains of tweets start there. Perhaps analyzing all the tweets of a single story over time to see how usage changes?